Skip to contentSkip to conversation

OpenRISC clone3 support

PATCH Superseded Stafford Horne 2 participants 12 messages Feb 26, 2020 – Feb 27, 2020 Openrisc architecture

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Stafford Horne

author maintainer
#0 lore raw

cover OpenRISC clone3 support

This series fixes the clone3 not implemented warnings I have been seeing during recent builds. It was a simple case of implementing copy_thread_tls and turning on clone3 generic support. Testing shows no issues.

Stafford Horne (3):
openrisc: Convert copy_thread to copy_thread_tls
openrisc: Enable the clone3 syscall
openrisc: Cleanup copy_thread_tls docs and comments

 arch/openrisc/Kconfig                   |  1 +
 arch/openrisc/include/uapi/asm/unistd.h |  1 +
 arch/openrisc/kernel/process.c          | 19 +++++++------------
 3 files changed, 9 insertions(+), 12 deletions(-)
Signature · 1 line
2.21.0
author maintainer
#1 lore raw

PATCH 1/3 openrisc: Convert copy_thread to copy_thread_tls

This is required for clone3 which passes the TLS value through a struct rather than a register.

---
 arch/openrisc/Kconfig          |  1 +
 arch/openrisc/kernel/process.c | 15 +++++----------
 2 files changed, 6 insertions(+), 10 deletions(-)
arch/openrisc/Kconfig +1 -0
index 1928e061ff96..5debdbe6fc35 100644
@@ -14,6 +14,7 @@ config OPENRISC
 	select HANDLE_DOMAIN_IRQ
 	select GPIOLIB
 	select HAVE_ARCH_TRACEHOOK
+	select HAVE_COPY_THREAD_TLS
 	select SPARSE_IRQ
 	select GENERIC_IRQ_CHIP
 	select GENERIC_IRQ_PROBE
arch/openrisc/kernel/process.c +5 -10
index b06f84f6676f..6695f167e126 100644
@@ -117,12 +117,13 @@ void release_thread(struct task_struct *dead_task)
 extern asmlinkage void ret_from_fork(void);
 
 /*
- * copy_thread
+ * copy_thread_tls
  * @clone_flags: flags
  * @usp: user stack pointer or fn for kernel thread
  * @arg: arg to fn for kernel thread; always NULL for userspace thread
  * @p: the newly created task
  * @regs: CPU context to copy for userspace thread; always NULL for kthread
+ * @tls: the Thread Local Storate pointer for the new process
  *
  * At the top of a newly initialized kernel stack are two stacked pt_reg
  * structures.  The first (topmost) is the userspace context of the thread.
@@ -148,8 +149,8 @@ extern asmlinkage void ret_from_fork(void);
  */
 
 int
-copy_thread(unsigned long clone_flags, unsigned long usp,
-	    unsigned long arg, struct task_struct *p)
+copy_thread_tls(unsigned long clone_flags, unsigned long usp,
+		unsigned long arg, struct task_struct *p, unsigned long tls)
 {
 	struct pt_regs *userregs;
 	struct pt_regs *kregs;
@@ -180,15 +181,9 @@ copy_thread(unsigned long clone_flags, unsigned long usp,
 
 		/*
 		 * For CLONE_SETTLS set "tp" (r10) to the TLS pointer passed to sys_clone.
-		 *
-		 * The kernel entry is:
-		 *	int clone (long flags, void *child_stack, int *parent_tid,
-		 *		int *child_tid, struct void *tls)
-		 *
-		 * This makes the source r7 in the kernel registers.
 		 */
 		if (clone_flags & CLONE_SETTLS)
-			userregs->gpr[10] = userregs->gpr[7];
+			userregs->gpr[10] = tls;
 
 		userregs->gpr[11] = 0;	/* Result from fork() */
 
Signature · 1 line
2.21.0
author maintainer
#2 lore raw

Re: [PATCH 1/3] openrisc: Convert copy_thread to copy_thread_tls

Quoted 2 lines · Stafford Horne

This is required for clone3 which passes the TLS value through a struct rather than a register.

[...]

Quoted 6 lines
arch/openrisc/kernel/process.c +0 -0
index b06f84f6676f..6695f167e126 100644
@@ -117,12 +117,13 @@ void release_thread(struct task_struct *dead_task)
 extern asmlinkage void ret_from_fork(void);

arch/openrisc/kernel/process.c void release_thread(struct task_struct *dead_task) · line 126

Quoted 10 lines
arch/openrisc/kernel/process.c +2 -1
 
 /*
- * copy_thread
+ * copy_thread_tls
  * @clone_flags: flags
  * @usp: user stack pointer or fn for kernel thread
  * @arg: arg to fn for kernel thread; always NULL for userspace thread
  * @p: the newly created task
  * @regs: CPU context to copy for userspace thread; always NULL for kthread
+ * @tls: the Thread Local Storate pointer for the new process

This should be *Storage*.

#3 lore raw

Re: [PATCH 1/3] openrisc: Convert copy_thread to copy_thread_tls

In reply to Stafford Horne
Quoted 52 lines · Stafford Horne

This is required for clone3 which passes the TLS value through a struct rather than a register.

---
 arch/openrisc/Kconfig          |  1 +
 arch/openrisc/kernel/process.c | 15 +++++----------
 2 files changed, 6 insertions(+), 10 deletions(-)
arch/openrisc/Kconfig +1 -0
index 1928e061ff96..5debdbe6fc35 100644
@@ -14,6 +14,7 @@ config OPENRISC
 	select HANDLE_DOMAIN_IRQ
 	select GPIOLIB
 	select HAVE_ARCH_TRACEHOOK
+	select HAVE_COPY_THREAD_TLS
 	select SPARSE_IRQ
 	select GENERIC_IRQ_CHIP
 	select GENERIC_IRQ_PROBE
arch/openrisc/kernel/process.c +4 -3
index b06f84f6676f..6695f167e126 100644
@@ -117,12 +117,13 @@ void release_thread(struct task_struct *dead_task)
 extern asmlinkage void ret_from_fork(void);
 
 /*
- * copy_thread
+ * copy_thread_tls
  * @clone_flags: flags
  * @usp: user stack pointer or fn for kernel thread
  * @arg: arg to fn for kernel thread; always NULL for userspace thread
  * @p: the newly created task
  * @regs: CPU context to copy for userspace thread; always NULL for kthread
+ * @tls: the Thread Local Storate pointer for the new process
  *
  * At the top of a newly initialized kernel stack are two stacked pt_reg
  * structures.  The first (topmost) is the userspace context of the thread.
@@ -148,8 +149,8 @@ extern asmlinkage void ret_from_fork(void);
  */
 
 int
-copy_thread(unsigned long clone_flags, unsigned long usp,
-	    unsigned long arg, struct task_struct *p)
+copy_thread_tls(unsigned long clone_flags, unsigned long usp,
+		unsigned long arg, struct task_struct *p, unsigned long tls)
 {
 	struct pt_regs *userregs;
 	struct pt_regs *kregs;

arch/openrisc/kernel/process.c copy_thread(unsigned long clone_flags, unsigned long usp, · line 183

Quoted 3 lines
arch/openrisc/kernel/process.c +0 -0
@@ -180,15 +181,9 @@ copy_thread(unsigned long clone_flags, unsigned long usp,
 
 		/*
 		 * For CLONE_SETTLS set "tp" (r10) to the TLS pointer passed to sys_clone.

Maybe reword this to:

For CLONE_SETTLS set "tp" (r10) to the TLS pointer. We probably shouldn't mention clone() explicitly anymore, since we now have clone3() and therefore two callers that pass in tls arguments.

Thanks!
Christian

author maintainer
#4 lore raw

Re: [PATCH 1/3] openrisc: Convert copy_thread to copy_thread_tls

Quoted 54 lines · Christian Brauner
Quoted 53 lines · Stafford Horne

This is required for clone3 which passes the TLS value through a struct rather than a register.

---
 arch/openrisc/Kconfig          |  1 +
 arch/openrisc/kernel/process.c | 15 +++++----------
 2 files changed, 6 insertions(+), 10 deletions(-)
arch/openrisc/Kconfig +1 -0
index 1928e061ff96..5debdbe6fc35 100644
@@ -14,6 +14,7 @@ config OPENRISC
 	select HANDLE_DOMAIN_IRQ
 	select GPIOLIB
 	select HAVE_ARCH_TRACEHOOK
+	select HAVE_COPY_THREAD_TLS
 	select SPARSE_IRQ
 	select GENERIC_IRQ_CHIP
 	select GENERIC_IRQ_PROBE
arch/openrisc/kernel/process.c +4 -3
index b06f84f6676f..6695f167e126 100644
@@ -117,12 +117,13 @@ void release_thread(struct task_struct *dead_task)
 extern asmlinkage void ret_from_fork(void);
 
 /*
- * copy_thread
+ * copy_thread_tls
  * @clone_flags: flags
  * @usp: user stack pointer or fn for kernel thread
  * @arg: arg to fn for kernel thread; always NULL for userspace thread
  * @p: the newly created task
  * @regs: CPU context to copy for userspace thread; always NULL for kthread
+ * @tls: the Thread Local Storate pointer for the new process
  *
  * At the top of a newly initialized kernel stack are two stacked pt_reg
  * structures.  The first (topmost) is the userspace context of the thread.
@@ -148,8 +149,8 @@ extern asmlinkage void ret_from_fork(void);
  */
 
 int
-copy_thread(unsigned long clone_flags, unsigned long usp,
-	    unsigned long arg, struct task_struct *p)
+copy_thread_tls(unsigned long clone_flags, unsigned long usp,
+		unsigned long arg, struct task_struct *p, unsigned long tls)
 {
 	struct pt_regs *userregs;
 	struct pt_regs *kregs;
@@ -180,15 +181,9 @@ copy_thread(unsigned long clone_flags, unsigned long usp,
 
Quoted 8 lines

arch/openrisc/kernel/process.c copy_thread(unsigned long clone_flags, unsigned long usp, · line 183

Quoted 2 lines
arch/openrisc/kernel/process.c +0 -0
 		/*
 		 * For CLONE_SETTLS set "tp" (r10) to the TLS pointer passed to sys_clone.

Maybe reword this to:

For CLONE_SETTLS set "tp" (r10) to the TLS pointer. We probably shouldn't mention clone() explicitly anymore, since we now have clone3() and therefore two callers that pass in tls arguments.

Sure, I updated it in the 'docs' commit, but as you mention I can just remove the mention of clone* all together. I will just remove that here and it won't have to be touched in the 'docs' commit.

author maintainer
#5 lore raw

PATCH 2/3 openrisc: Enable the clone3 syscall

In reply to Stafford Horne

Enable the clone3 syscall for OpenRISC. We use the generic version.

This was tested with the clone3 test from selftests. Note, for all tests to pass it required enabling CONFIG_NAMESPACES which is not enabled in the default kernel config.

---
 arch/openrisc/include/uapi/asm/unistd.h | 1 +
 1 file changed, 1 insertion(+)
arch/openrisc/include/uapi/asm/unistd.h +1 -0
index 566f8c4f8047..fae34c60fa88 100644
@@ -24,6 +24,7 @@
 #define __ARCH_WANT_SET_GET_RLIMIT
 #define __ARCH_WANT_SYS_FORK
 #define __ARCH_WANT_SYS_CLONE
+#define __ARCH_WANT_SYS_CLONE3
 #define __ARCH_WANT_TIME32_SYSCALLS
 
 #include <asm-generic/unistd.h>
Signature · 1 line
2.21.0
Acked-by on v1 2/3
#6 lore raw

Re: [PATCH 2/3] openrisc: Enable the clone3 syscall

Quoted 5 lines · Stafford Horne

Enable the clone3 syscall for OpenRISC. We use the generic version.

This was tested with the clone3 test from selftests. Note, for all tests to pass it required enabling CONFIG_NAMESPACES which is not enabled in the default kernel config.

For OpenRISC, I assume. Hm, maybe we should fix the tests to skip when CONFIG_NAMESPACES is not enabled.

Quoted 2 lines
author maintainer
#7 lore raw

Re: [PATCH 2/3] openrisc: Enable the clone3 syscall

Quoted 9 lines · Christian Brauner
Quoted 5 lines · Stafford Horne

Enable the clone3 syscall for OpenRISC. We use the generic version.

This was tested with the clone3 test from selftests. Note, for all tests to pass it required enabling CONFIG_NAMESPACES which is not enabled in the default kernel config.

For OpenRISC, I assume. Hm, maybe we should fix the tests to skip when CONFIG_NAMESPACES is not enabled.

Yes, not the default for openrisc defconfig. It might make sense to either skip the tests of have them as expected fails when CONFIG_NAMESPACES is off.

On the otherhand, I am not sure if the self tests know about the CONFIG_* available. I notice many test directories have a 'config' file and a readme saying, for these test to run ensure use have at least 'these' config values set.

-Stafford

author maintainer
#8 lore raw

PATCH 3/3 openrisc: Cleanup copy_thread_tls docs and comments

In reply to Stafford Horne

Previously copy_thread_tls was copy_thread and before that something else. Remove the documentation about the regs parameter that didn't exist in either version.

Next, fix comment wrapping and details about how TLS pointer gets to the copy_thread_tls function.

---
 arch/openrisc/kernel/process.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/openrisc/kernel/process.c +2 -2
index 6695f167e126..b442e7b59e17 100644
@@ -122,7 +122,6 @@ extern asmlinkage void ret_from_fork(void);
  * @usp: user stack pointer or fn for kernel thread
  * @arg: arg to fn for kernel thread; always NULL for userspace thread
  * @p: the newly created task
- * @regs: CPU context to copy for userspace thread; always NULL for kthread
  * @tls: the Thread Local Storate pointer for the new process
  *
  * At the top of a newly initialized kernel stack are two stacked pt_reg
@@ -180,7 +179,8 @@ copy_thread_tls(unsigned long clone_flags, unsigned long usp,
 			userregs->sp = usp;
 
 		/*
-		 * For CLONE_SETTLS set "tp" (r10) to the TLS pointer passed to sys_clone.
+		 * For CLONE_SETTLS set "tp" (r10) to the TLS pointer passed
+		 * in clone_args to sys_clone3.
 		 */
 		if (clone_flags & CLONE_SETTLS)
 			userregs->gpr[10] = tls;
Signature · 1 line
2.21.0
Acked-by on v1 3/3
#9 lore raw

Re: [PATCH 3/3] openrisc: Cleanup copy_thread_tls docs and comments

Quoted 8 lines · Stafford Horne

Previously copy_thread_tls was copy_thread and before that something else. Remove the documentation about the regs parameter that didn't exist in either version.

Next, fix comment wrapping and details about how TLS pointer gets to the copy_thread_tls function.

Quoted 17 lines
---
 arch/openrisc/kernel/process.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/openrisc/kernel/process.c +0 -1
index 6695f167e126..b442e7b59e17 100644
@@ -122,7 +122,6 @@ extern asmlinkage void ret_from_fork(void);
  * @usp: user stack pointer or fn for kernel thread
  * @arg: arg to fn for kernel thread; always NULL for userspace thread
  * @p: the newly created task
- * @regs: CPU context to copy for userspace thread; always NULL for kthread
  * @tls: the Thread Local Storate pointer for the new process
  *
  * At the top of a newly initialized kernel stack are two stacked pt_reg

arch/openrisc/kernel/process.c copy_thread_tls(unsigned long clone_flags, unsigned long usp, · line 183

Quoted 6 lines
arch/openrisc/kernel/process.c +2 -1
@@ -180,7 +179,8 @@ copy_thread_tls(unsigned long clone_flags, unsigned long usp,
 			userregs->sp = usp;
 
 		/*
-		 * For CLONE_SETTLS set "tp" (r10) to the TLS pointer passed to sys_clone.
+		 * For CLONE_SETTLS set "tp" (r10) to the TLS pointer passed
+		 * in clone_args to sys_clone3.

As I said in my other reply, I'd not reference any specific caller since we have at least two.

Christian

#10 lore raw
In reply to Stafford Horne
Quoted 3 lines · Stafford Horne

This series fixes the clone3 not implemented warnings I have been seeing during recent builds. It was a simple case of implementing copy_thread_tls and turning on clone3 generic support. Testing shows no issues.

This all looks good to me. Thanks for doing this. We're getting closer and closer to having all architectures supporting clone3()!

You want me to pick this series up for 5.7 or are you going through another tree?

Christian

author maintainer
#11 lore raw
Quoted 10 lines · Christian Brauner
Quoted 3 lines · Stafford Horne

This series fixes the clone3 not implemented warnings I have been seeing during recent builds. It was a simple case of implementing copy_thread_tls and turning on clone3 generic support. Testing shows no issues.

This all looks good to me. Thanks for doing this. We're getting closer and closer to having all architectures supporting clone3()!

You want me to pick this series up for 5.7 or are you going through another tree?

I am the OpenRISC maintainer so I will just send these through my tree during the 5.7 merge window with you Ack attributions.

-Stafford